Dysfunction of Endothelial Protein C Activation in ...

The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

DYSFUNCTION OF ENDOTHELIAL PROTEIN C ACTIVATION IN SEVERE MENINGOCOCCAL SEPSIS SAUL N. FAUST, M.R.C.P., MICHAEL LEVIN, F.R.C.P., PH.D., ODILE B. HARRISON, B.SC., ROBERT D. GOLDIN, F.R.C.PATH., MARION S. LOCKHART, B.SC., SHEILA KONDAVEETI, PH.D., ZOLTAN LASZIK, M.D., CHARLES T. ESMON, PH.D., AND ROBERT S. HEYDERMAN, M.R.C.P., PH.D.

ABSTRACT Background Impairment of the protein C anticoagulation pathway is critical to the thrombosis associated with sepsis and to the development of purpura fulminans in meningococcemia. We studied the expression of thrombomodulin and the endothelial protein C receptor in the dermal microvasculature of children with severe meningococcemia and purpuric or petechial lesions. Methods We assessed the integrity of the endothelium and the expression of thrombomodulin and the endothelial protein C receptor in biopsy specimens of purpuric lesions from 21 children with meningococcal sepsis (median age, 41 months), as compared with control skin-biopsy specimens. Results The expression of endothelial thrombomodulin and of the endothelial protein C receptor was lower in the patients with meningococcal sepsis than in the controls, both in vessels with thrombosis and in vessels without thrombosis. On electron microscopical examination, the endothelial cells were generally intact in both thrombosed and nonthrombosed vessels. Plasma thrombomodulin levels in the children with meningococcal sepsis (median, 6.4 ng per liter) were higher than those in the controls (median, 3.6 ng per liter; P=0.002). Plasma levels of protein C antigen, protein S antigen, and antithrombin antigen were lower than those in the controls. In two patients treated with unactivated protein C concentrate, activated protein C was undetectable at the time of admission, and plasma levels remained low. Conclusions In severe meningococcal sepsis, protein C activation is impaired, a finding consistent with down-regulation of the endothelial thrombomodulin– endothelial protein C receptor pathway. (N Engl J Med 2001;345:408-16.) Copyright © 2001 Massachusetts Medical Society.

N

EISSERIA MENINGITIDIS is the leading infectious cause of death in children in developed countries and is a cause of disability resulting from extensive skin damage and loss of limbs.1,2 Severe meningococcal sepsis is characterized by marked inflammatorycell activation, disseminated intravascular coagulation, and vascular compromise.3-5 As compared with other forms of septic shock, the coagulopathy and microvascular thrombosis that develop in this type of sepsis are particularly severe. Purpura fulminans occurs in 10 to 20 percent of cases6,7 and in severe cases involves thrombosis of the large vessels with infarction of the

digits and limbs.6,8 This disorder results from complex dysregulation of normal hemostatic mechanisms.6-8 Procoagulant pathways are activated,9-12 and there is impairment of both the natural anticoagulant pathways13-17 and the fibrinolytic system.18-21 It remains unclear why purpura fulminans develops in some patients, whereas in others with equally severe septic shock there are no thrombotic complications. Dysfunction of the protein C activation pathway appears to be critical to the development of thrombosis in purpura fulminans.6,7,22 Protein C is a vitamin K–dependent glycoprotein that circulates in plasma as an inactive zymogen. Once activated, protein C requires protein S as a cofactor for its anticoagulant functions. Congenital and acquired deficiencies of protein C or of protein S may result in purpura fulminans.22-24 Replacement therapy with protein C concentrate prevents purpura fulminans in children with a congenital deficiency of this glycoprotein, and infusions of activated protein C have been shown to moderate the development of coagulopathy and prevent death in animal models of gram-negative sepsis and in humans with severe sepsis.25,26 In meningococcal disease, plasma levels of protein C and protein S are markedly reduced,13,16,17,27 but dysfunction of the endothelial protein C activation pathway may also be involved. Activation of protein C requires binding of the protein to two receptors on the endothelial surface: thrombomodulin and the endothelial protein C receptor.22 The resulting complex acts as a molecular switch that limits the procoagulant activity of thrombin (Fig. 1); this complex also has a number of antiinflammatory properties.22 We postulated that disruption of the activated endothelial protein C complex is an early event in the development of the widespread thrombosis and disseminated intravascular coagulation associated with severe meningococcal disease. To test this hypothesis, we studied the expression of thrombomodulin and the endothelial protein C From the Departments of Paediatrics (S.N.F., M.L., O.B.H., S.K.) and Pathology (R.D.G.), Imperial College School of Medicine at St. Mary’s Hospital, London; the Oklahoma Medical Research Foundation, Oklahoma City (M.S.L., C.T.E.); the Departments of Pathology (Z.L., C.T.E.) and Biochemistry and Molecular Biology (C.T.E.), University of Oklahoma Health Sciences Center, Oklahoma City; the Howard Hughes Medical Institute, Oklahoma City (C.T.E.); and the Departments of Pathology and Microbiology, University of Bristol, Bristol, United Kingdom (R.S.H.). Address reprint requests to Dr. Levin at the Department of Paediatrics, Imperial College School of Medicine at St. Mary’s Hospital, Norfolk Pl., London W2 1PG, United Kingdom, or at [email protected].

408 · N Engl J Med, Vol. 345, No. 6 · August 9, 2001 · www.nejm.org The New England Journal of Medicine Downloaded from nejm.org on September 16, 2015. For personal use only. No other uses without permission. Copyright © 2001 Massachusetts Medical Society. All rights reserved.

DYSF UNC T ION OF E NDOT HE L IAL P ROTEIN C AC TIVATION IN S EV ERE MENINGOC OCC A L SEPS IS

Vessel lumen

Thrombin Xa

Xa

Protein C

Tissue factor Complex 2

Inactive cofactors

V

Vlll

Complex 3

Active cofactors

Endothelial membrane

Activated protein C

V

Viii Complex 1

Cofactor inactivation

Protein C activation

Thrombin generation

Initiation of coagulation

Thrombomodulin

Endothelial protein C receptor

Protein S

Cytoplasm Figure 1. Endothelial Activation of Coagulation and the Protein C Pathway. Coagulation is initiated by tissue factor and other coagulation-factor complexes on the surface of endothelial cells and monocytes. The activated factor X that is consequently generated requires activated cofactors V and VIII to produce thrombin, which in turn forms a complex with thrombomodulin. Protein C activation takes place by way of interaction between the thrombomodulin–thrombin complex and the endothelial protein C receptor. Activated protein C, together with its cofactor, protein S, inactivates factors V and VIII to provide negative feedback to the generation of thrombin. Complex 1 comprises tissue factor and coagulation factors VII, IX, and X; complex 2 comprises factors IX and X and cofactor VIII; and complex 3 comprises factor X, prothrombin, and cofactor V.

receptor on the dermal endothelium in skin-biopsy specimens from children with severe meningococcal disease and related these findings to events in the circulation. METHODS Patients and Specimens Meningococcal disease was diagnosed according to clinical criteria and was confirmed microbiologically in children admitted to St. Mary’s Hospital, London.19,28 The study was approved by the St. Mary’s local research-ethics committee and was conducted between January 1998 and November 2000. Informed consent was obtained from the parents of the hospitalized children for the collection of the clinical samples and from the healthy adults and the parents of the healthy children who served as controls. Plasma samples were obtained from 83 children with meningococcal disease at recorded intervals after the first administration of parenteral antibiotics. The median age of the children was 37 months (range, 1 to 210), and the median Glasgow Meningococcal Septicemia Prognostic Score was 12 (range, 5 to 15) on a scale of 0 to 15 (higher scores denote more severe disease).29 Two patients received treatment with unactivated protein C concentrate. The patients were categorized according to the clinical severity of their cutaneous disease, as follows: no scarring or only mild disease, severe scarring (no plastic surgery required), or purpura fulminans (requiring amputation or skin grafting or resulting in death due

to severe disease). Control plasma samples were obtained from eight adults and eight healthy children before routine surgery. Specimens of skin 3 mm in diameter were obtained by punch biopsy from the edge of purpuric or petechial lesions in 21 patients with severe meningococcal sepsis (a consecutive subgroup of the cohort of 83 children).30 The median age of these 21 children was 41 months (range, 1 to 185). The median Glasgow Meningococcal Septicemia Prognostic Score was 11 of 15 (range, 6 to 15), and 9 of 15 patients had a Pediatric Risk of Mortality score above 50 percent.31 All the biopsy specimens were taken within 24 hours after the administration of the first dose of parenteral antibiotics. In two additional patients, biopsy specimens were also obtained three months into recovery, during plastic-surgery procedures. Control skin-biopsy specimens were obtained from five children during routine surgery. Immunohistochemical Studies Formalin-fixed, paraffin-embedded sections of the skin-biopsy specimens were immunostained for thrombomodulin, endothelial protein C receptor, a neutrophil marker (neutrophil elastase), endothelial-cell markers (CD31 and CD34), and a monocyte marker (CD68) by the avidin–biotin–peroxidase method.32 Sections 3 µm thick were incubated with monoclonal antibodies to detect expression of thrombomodulin (antibody, 7.8 µg per milliliter), endothelial protein C receptor (antibody, 0.2 mg per milliliter),33 neutrophil elastase (M0752, Dako), CD31 (M0823, Dako), CD34 (M7165, Dako), and CD68 (M0876, Dako). Antigen retrieval by

N Engl J Med, Vol. 345, No. 6 · August 9, 2001 · www.nejm.org · 409 The New England Journal of Medicine Downloaded from nejm.org on September 16, 2015. For personal use only. No other uses without permission. Copyright © 2001 Massachusetts Medical Society. All rights reserved.


microwave heat induction in citrate buffer (pH 6.0; HDS05, SD Supplies) was required for optimal staining with the anti-CD31, anti-CD34, and anti-CD68 antibodies.34 Primary antibody binding was detected with an immunoperoxidase kit (Vectastain Elite ABC kit, Vector Laboratories). In each section, the degree of thrombosis was assessed (severe [more than 66 percent of the vessels thrombosed], moderate [33 to 66 percent of the vessels thrombosed], or mild [less than 33 percent of the vessels thrombosed]), as was the degree of inflammation (severe, moderate, or mild). The intensity of immunostaining was graded semiquantitatively as strong, moderate, weak, or absent by comparison with the staining observed in control skin specimens, with strong staining equivalent to that in the controls. For each antigen, two to five sections, with 30 to 75 µm between sections, were studied to ensure that in all the biopsy specimens, both normal and abnormal tissue was obtained. To avoid bias, all the sections were initially compared with the controls by a single investigator, who was unaware of the identity of the patients and the severity of their illness. For each antigen, 20 random sections were then assessed separately by two independent, blinded investigators to ensure that the interpretation of the staining was consistent and accurate. In addition, 15 biopsy specimens were examined by two independent histopathologists who had no knowledge of the nature of the study. Additional information is in Supplementary Appendix 1 (available with the full text of this article at http://www.nejm.org).

A

B

C

Electron Microscopy For electron microscopy, skin-biopsy specimens were fixed in 4 percent glutaraldehyde and stained with osmium tetroxide. Final staining was with uranyl acetate and Reynold’s lead citrate. Transmission electron microscopy (model 400, Philips) was carried out at magnifications of 1000 to 2800. Assays of Plasma Antigens Thrombomodulin antigen and thrombin–antithrombin complexes were detected with the use of a thrombomodulin enzymelinked immunosorbent assay (ELISA) kit (Immubind [no. 837], American Diagnostica) and a TAT-complex ELISA kit (Enzygnost, Dade Behring). Endothelial protein C receptor, protein C, protein S, and antithrombin antigens were measured by ELISA.35-38 Activated protein C in plasma was measured with the use of a modified enzyme-capture assay.39 The lower limit of detection was 3 ng per milliliter. Details of this method are in Supplementary Appendix 2 (available with the full text of this article at http://www. nejm.org). Statistical Analysis Data are expressed as medians, means, and ranges. Comparisons between the patients and the controls were performed with use of paired t-tests with log-transformed data.

RESULTS Histologic and Ultrastructural Findings

In all the biopsy specimens, the general tissue structure was well preserved; there was evidence of thrombosis and frequently of a perivascular, acute inflammatory-cell infiltrate of neutrophils and monocytes (Fig. 2). Thrombosis was classified as severe in 5 of the 21 patients, moderate in 9, mild in 4, and absent in 3. Inflammation was assessed as severe in 6 of these 21 patients, moderate in 12, and mild in 3. The intensity of staining for the endothelial-cell markers CD31 and CD34 in the biopsy specimens from the patients with meningococcal sepsis was generally equivalent to that observed in the control skin spec-

Figure 2. Skin-Biopsy Specimen from a Patient with Meningococcal Sepsis. A biopsy specimen from a purpuric lesion shows areas containing thrombosed vessels (right-hand arrow in Panel A and arrow in Panel B) and a perivascular infiltrate (left-hand arrow in Panel A) (hematoxylin and eosin, ¬100 [Panel A] and ¬400 [Panel B]). Panel C shows inflammatory cells (arrow) around nonthrombosed vessels (hematoxylin and eosin, ¬400). The cellular infiltrate consisted of both neutrophils (identified by neutrophil-elastase staining) and monocytes and macrophages (identified by CD68 staining).

imens, but discontinuous staining was occasionally observed in both thrombosed and nonthrombosed vessels. Transmission electron microscopy of skin-biopsy specimens from five children with purpuric lesions revealed that although some vessels (both those with and those without thrombosis) showed loss of endothelial cells with a characteristic lack of organelles, this finding was not widespread. The integrity of the endothelial cells was generally preserved in both thrombosed and nonthrombosed vessels (Fig. 3). Thrombomodulin and Endothelial Protein C Receptor Expression on Endothelium

The intensity of staining for endothelial thrombomodulin in the biopsy specimens from all the patients with meningococcal sepsis was less than that in the specimens from the controls (Fig. 4). Thrombomodulin staining was weak in 15 and moderate in 6 of the 21 specimens from the patients with menin-

410 · N Engl J Med, Vol. 345, No. 6 · August 9, 2001 · www.nejm.org The New England Journal of Medicine Downloaded from nejm.org on September 16, 2015. For personal use only. No other uses without permission. Copyright © 2001 Massachusetts Medical Society. All rights reserved.

DYSF UNC T ION OF E NDOT HE L IAL P ROTEIN C AC TIVATION IN S EV ERE MENINGOC OCC A L SEPS IS

A

B

Figure 3. Transmission Electron Micrographs of Skin-Biopsy Specimens from Patients with Meningococcal Sepsis. The endothelium is intact (arrows) in both a thrombosed vessel (Panel A, ¬1300) and a nonthrombosed vessel (Panel B, ¬2200). The black areas are artifacts that result from the processing of small and fragile specimens.

gococcal sepsis. The intensity of staining for endothelial protein C receptor was lower in 17 of these 21 specimens than in the control specimens: it was judged to be weak in 11 and moderate in 6, but staining was strong (i.e., equivalent to that of the controls) in the other 4 specimens (Fig. 5). Comparatively weak staining for thrombomodulin and endothelial protein C receptor was also observed in areas with little thrombosis. Occasionally, areas of weak thrombomodulin staining were identified where moderate or strong staining for endothelial protein C receptor was preserved in adjacent sections. In two patients

with severe purpura fulminans, comparative examination of skin-biopsy specimens over time revealed the resolution of most of the initial changes, with some residual acute inflammatory infiltrate three months later (Fig. 6). Plasma Levels of Thrombomodulin, Endothelial Protein C Receptor, Protein C, Protein S, and Antithrombin Antigens and Thrombin–Antithrombin Complexes

The plasma levels of thrombomodulin antigen were significantly higher in all the patients with meningococcal disease on day 1 after the initiation of treat-

N Engl J Med, Vol. 345, No. 6 · August 9, 2001 · www.nejm.org · 411 The New England Journal of Medicine Downloaded from nejm.org on September 16, 2015. For personal use only. No other uses without permission. Copyright © 2001 Massachusetts Medical Society. All rights reserved.


A

A

C

B

B

D

D

Figure 4. Skin-Biopsy Specimens Incubated with a Monoclonal Antibody against Thrombomodulin (Immunoperoxidase Stain). Panel A (¬200) shows normal skin with intense thrombomodulin staining on endothelial cells (arrow). Epidermal cells are known to stain nonspecifically with this antibody33 (as seen in the area of dense staining [upper right]). Panels B (¬200), C (¬400), and D (¬200) show skin specimens from patients with meningococcal sepsis. There is reduced thrombomodulin staining in nonthrombosed vessels (arrows in Panels B and D) and thrombosed vessels (arrows in Panel C).

ment than in the controls (P=0.002). The levels had returned to normal six to eight weeks later. The levels of thrombomodulin antigen correlated with the severity of disease (Table 1). Plasma levels of endothelial protein C receptor antigen in the children with moderate disease were not significantly different from those in either the children or the adults who served as controls (P=0.35 and P=0.30, respectively). Plasma levels of protein C, protein S, and antithrombin antigens were also low on day 1 (P